Research Final

Front 1

Descriptive statistics

Back 1

describe what the data shows

numeric values that help summarize a dataset to describe basic features of the data

does NOT attempt to infer a "larger truth"

Front 2

Inferential Statistics

Back 2

applying probability theory and math models to a set of data to reach conclusions that extend beyond the immediate data

try to reach conclusions beyond the data that you have - to 'infer" something about the general population

Front 3

purposes of inferential statistics

Back 3

estimation

hypothesis testing

Front 4

types of descriptive statistics

Back 4

measures of central tendency

measures of dispersion

describing distributions

Front 5

measures of central tendency

Back 5

mean

median

mode

Front 6

Mean

Back 6

math average

most often used - best if data are normally distributed

Front 7

median

Back 7

middle value in an ordered set - puth them in order from low to high and see what number is in the exact middle

often used if data is skewed - not a normal distribution (bell shaped curve)

Front 8

mode

Back 8

value that occurs most often in your data set - may be used when the data are normal OR skewed

Front 9

types of measures of dispersion

Back 9

range

variance

standard deviation

Front 10

dispersion

Back 10

spread of values around the central tendency - how different are the values from each other?

Front 11

range

Back 11

maximum - minimum

gross measurement - one single outlier can exaggerate the range

Front 12

variance

Back 12

sum of squares / # values - 1

take each value and subtract the mean then square the value - add all the squares together then take the number of values and subtract to find the number to divide by

this is the standard deviation squared

Front 13

standard deviation

Back 13

more accurate and detailed estimate of dispersion - show relation that a set of scores has to the mean in units that are meaningful

square root of the variance

Front 14

standard deviation and % values

Back 14

1 SD = 68% of scores
2 SD = 95% of scores
3 SD = 99% of scores

Front 15

normally distributed data

Back 15

then every number of central tendency will be about the same (i.e.: mean, median, and mode will all be approximately the same number)

Front 16

describing distributions

Back 16

shape

skewness

kurtosis

Front 17

shape

Back 17

symmetrical vs asymmetrical

if you fold it in half does it look the same?

Front 18

modality

Back 18

unimodal - 1 hump

bimodal - 2 humps

trimodal - 3 humps

easiest number to see is the mode on the graph because it will have the most occurrences

Front 19

skewness

Back 19

indicates the nature of asymmetrical distributions

if the data is skewed then your mean, median, and mode are not the same

if the data are skewed then the standard deviation percentages do not apply

Front 20

positive skewness

Back 20

the data are predominately located to the left side of the graph

Front 21

negative skewness

Back 21

the data are predominately located to the right side of the graph

Front 22

kurtosis

Back 22

the measure of how peaked or flat the distribution is

helps to describe normal distribution

Front 23

leptokurtic

Back 23

high kurtosis -

leaping up - very peaked

all values are very similar and close to the mean

Front 24

platykurtic

Back 24

low kurtosis

flat - it is flatter in data points

still with normal distribution but flatter distribution

Front 25

purpose of results section

Back 25

objectively present key findings from your study

uses order and logic
typically includes tables
descriptive and inferential stats
summarize results but don't interpret them until discussion

Front 26

sample

Back 26

just one of infinite number o possible samples from the population of interest

Front 27

sampling distribution

Back 27

normal distribution even if the data from each sample is NOT normally distributed

Front 28

what descriptive statistic would give you the best guess at the population mean?

Back 28

the sample mean

the larger the sample the better the estimate for the population mean

Front 29

what descriptive statistic from the sample would allow you to know how likely your sample mean reflects your population mean?

Back 29

standard deviation - tells you how much fluctuation there is -

smaller standard deviation means you are more confident that the average (mean) is correct for the population mean

Front 30

standard error of the mean

Back 30

theoretic value - reflects the sampling distribution

comparable to the standard deviation of raw scores except it relates to the distribution of sample means

SEM = SD/ square root of N

Front 31

How do you get a small Standard Error of the Mean (SEM)

Back 31

large population (N)

small standard deviation - more leptokurtic distribution of raw values

Front 32

How does the standard error of the mean compare to the standard deviation?

Back 32

The standard error of the mean is smaller than the standard deviation - by definition this must be true because the SEM is the SD / square root of N

The sampling distribution will be more leptokurtic than the raw data

Front 33

2 purposes of inferential statistics

Back 33

estimation - estimate information about the population - how confident we are in the estimation (i.e.: mean and 95% confidence)

hypothesis testing - does data support the null hypothesis? (t tests, chi square, ANOVA, correlation coefficients, regression analysis)

Front 34

what is the confidence interval?

Back 34

the window that the true population mean (that we cannot actually measure and we use the sampling mean to estimate it) actually falls

Front 35

Inferential statistics allow the researcher to ...

Back 35

1 - calculate confidence intervals regarding the "true" population mean

2 - examine the probability of a given result assuming the null is true

Front 36

probability of observed values for inferential statistics

Back 36

these values are compared to known distributions of the value if the null was true; this allows us to determine what the probability is of obtaining that value assuming that the null hypothesis is true (the p value)

the higher the observed statistic, the more unlikely it is that it could occur by chance factors alone

Front 37

alpha

Back 37

a "p" value chosen by the researcher to determine significance level

typically set to 0.05 or 0.01 at the beginning of the study - not set later but from the very beginning

Front 38

statistically significant alpha

Back 38

saying statistically significant means that you reject the null hypothesis thereby accepting the alternative hypothesis

Front 39

alpha 0.01 vs 0.05

Back 39

0.01 more stringent than 0.05 - so you would be more confident that there is an effect than just by chance if you choose 0.01

therefore making a type 1 error

Front 40

statistical vs clinical significance

Back 40

statistical significance - the result it real - it is unlikely due to chance or sampling error -BUT - real differences aren't necessarily important - clinical significance is a judgment call and means that the differences observed are meaningful

Front 41

type 1 error

Back 41

we reject the null when it is actually true and we should have accepted it -
see something but it's not really there

"convict an innocent man"

Front 42

type 2 error

Back 42

we retain the null (fail to reject it) when it is actually false

should have seen something that was there but you didn't

"failure to convict a guilty man"

Front 43

p value

Back 43

the probability of getting the value if the null is true

this is VERY IMPORTANT to researchers

how you determine if something is statistically significant

Front 44

p = 0.87

Back 44

this means that 87% chance that it is related to the hypothesis testing

you want a low p value to know that the null is unlikely to be true because you conduct a test to know if the alternative hypothesis is true

Front 45

probability theory

Back 45

helps determine whether an observed statistic would be likely if the null hypothesis was true

the larger the observed statistic the more unlikely it is that such a result would be achieved by chance if the null was true

Front 46

types of T tests

Back 46

one sample t test

independent t test

paired samples t test

Front 47

T test

Back 47

inferential statistical test used to examine whether 2 groups (defined by nominal variable) are different on some dependent variable that is measured on an interval or ratio scale

ie: Do PA and AFN students differ in their average age?

IV = program
IV levels = 2, PA or AFN
DV = average age
DV type = ratio b/c 50 is twice as old as 25

Front 48

one sample t test

Back 48

comparing one group to a KNOWN population mean

Front 49

t test formula

Back 49

t= (mean of sample) - (population mean) / standard error of the mean

Front 50

how do you know if there is a large difference based on t?

Back 50

t would be around zero if there is no difference and the more it differs from zero the more you have a difference

Front 51

independent samples t test

Back 51

comparing one group to a separate group - there is not a KNOWN population mean for comparison -

ie: do men and women CRNAs earn different starting salaries? -

the differences are not known and published - only the total mean score is published / KNOWN data

Front 52

independent samples t test formula

Back 52

t = (mean group 1 - mean group 2) divided by pooled estimated standard error

Front 53

paired samples t test

Back 53

if we are comparing same group at two different times (i.e.: midterm and final - every student takes both exams - do they score differently?)

Front 54

chi square test of independence

Back 54

used to analyze relationship between 2 variables when both are nominal or ordinal (or when scores have been reduced to nominal or ordinal levels because they are not normally distributed in the sample) and the two variables are measured on the same individuals

Front 55

2x3 chi square analysis

Back 55

there are 6 possibilities of where you might fall

Front 56

if alpha = 0.01 and p = 0.011 is it significant

Back 56

no - the p is more than the alpha

Front 57

standard deviation - is it descriptive or inferential statistic?

Back 57

descriptive

Front 58

standard error of the mean - is it descriptive or inferential

Back 58

inferential

Front 59

what is the only value that tells you something is statistically significant?

Back 59

P value

Front 60

what values come from which tests?

Back 60

t = T test
r = correlational test
f = ANOVA test
x = chi square test

Front 61

what two test together tell you the probability that the null is true?

Back 61

the p value in conjunction with the alpha value

Front 62

p < or = 0.05 (assuming alpha = 0.05)

Back 62

conclude that because it is SO unlikely that the null is true we will assume it isn't - reject the null and retain the alternative

Front 63

bivariate linear regression

Back 63

common with pearson correlation - look at the same relationship but tells you an equation to predict one variable from another variable -

you have only one independent variable predicting only one dependent variable

Front 64

multiple regression

Back 64

dependent variable is what you are predicting and the independent variable is the predictor -

tells you if your prediction is statistically significant - is it valid - can you use it?

Front 65

p > alpha in multiple regression

Back 65

tells you independent variables are not good at predicting the dependent variable

if p > alpha you reject null

Front 66

one way ANOVA

Back 66

comparing the means of 3 or more groups

Front 67

factorial ANOVA

Back 67

2 or more independent variables - how many variables there are - not how many levels of the IV

Front 68

2 variables not related but want to see how they affect each other

Back 68

chi square test

Front 69

signal

Back 69

the amount of differences between groups

Front 70

noise

Back 70

the amount of differences between individuals in the same group

Front 71

the larger the inferential statistic the smaller the p value - is this good or bad and why

Back 71

good - you hope to reject the null and accept the alternative and a small p value allows you to do so with statistical significance

Front 72

how do you maximize the signal to noise ratio?

signal : noise

Back 72

use well selected, reliable and valid instruments to measure construct of interest

Front 73

why should you maximize N?

Back 73

helps to decrease the P value and get better determination of statistical significance -

use the largest you can afford and is permitted to you

Front 74

bivariate correlation

Back 74

the statistical index of the degree to which two variables are associated

"pearson correlation coefficient" (r)

Front 75

positive correlation coefficient

Back 75

values on one variable increase the values on the other variable also increase

r = up to +1

Front 76

negative correlation coefficient

Back 76

values on one variable increase then the values on the other variable decrease

r = 0 to -1

Front 77

strength or magnitude of correlation coefficient

Back 77

numerical value between -1 and 1

closer to 1 is stronger correlation and + or - tells you if they are associated or not associated

Front 78

coefficient of determination

Back 78

r2 (r squared)

indicates the percent of shard variance between 2 variables

describes the over lab between variables

Front 79

what is regression?

Back 79

extension of correlation - specifies IV versus DV - doesn't mean that one causes the other but provides a prediction equation and can simultaneously examine more than on X as a predictor of Y - so not limited to just 2 variables

Front 80

define regression

Back 80

statistical technique for finding a straight line that best predicts the relationship between variables

Front 81

y = b0 + b1x

Back 81

equation for bivariate linear regression

Front 82

what does y in bivariate linear regression equation mean?

Back 82

y is the predicted value of the y axis variable = the outcome or the Dependent Variable

Front 83

what does x in bivariate linear regression equation mean?

Back 83

x is the x axis variable - the predictor or Independent Variable

Front 84

b0 in the bivariate linear regression equation

Back 84

the intercept = value of y when x = 0

Front 85

what does b1 in the bivariate linear regression equation mean?

Back 85

the slop of the line = the amount of change in Y for each unit change in X

Front 86

what is the regression coefficient?

Back 86

b1 and it provides an index of the relationship between X and Y

it is the slope!

Front 87

h0: b1 = 0

Back 87

there is no linear relationship between x and y

Front 88

h0 : b1 not equal 0

Back 88

there is a linear relationship between X and Y

Front 89

what is the purpose of multiple linear regression?

Back 89

the use of several IV to predict a DV

Front 90

what do you need for each predictor in the multiple regression equation?

Back 90

there will be a null and alternative hypothesis for each indicator in the multiple regression equation

Front 91

what is the overall regression equation?

Back 91

R2 (capital R squared) - proportion of variability in our outcome that can predicted by the collection of predictors in the model - it includes each unique predictors contribution as well as the overlap contribution from each predictor

Front 92

what is a Factor?

Back 92

the independent variable that designates the group - this is part of the ANOVA terminology

treatment condition
year of high school
ethnicity

Front 93

what is a level?

Back 93

individual conditions or values that make up a factor

ie: drug a, b, or c
freshman, sophomore, junior
caucasian, asian, hispanic, etc..

Front 94

what is ANOVA? Why is it beneficial?

Back 94

analysis of variance - a hypothesis testing procedure used to examine mean differences between multiple (3+) groups

more efficient that multiple t tests

avoids complications with finding effects that aren't really there due to use of multiple tests (decreases type 1 error rates)

provides greater flexibility in designing experiments

Front 95

what does "p" represent?

Back 95

the probability that differences are due to chance

Front 96

what does "F" represent?

Back 96

the ratio of between groups to within groups variance

signal : noise

Front 97

what does if mean if F is small?

Back 97

the variability between groups (signal) is negligible compared to the variation within the group (noise)

F test non significant if p > 0.05

if F is small then p will be large and therefore not significant

Front 98

what does it mean if F is large?

Back 98

can retain the alternative hypothesis and reject the null hypothesis -

the variability between groups is large compared to the variation within the groups (noise)

F test will be significant and p < or = 0.05

Front 99

if F is significant then what?

Back 99

you know that there is a difference between the means but you don't know where the differences exist - so you need to follow up with further testing to tell where the means differ

post-hoc contrasts
planned comparisons

Front 100

what are effect size, relative risk, and odds ratio

Back 100

additional statistical data that are increasing in use in biomedical research - in addition to the other 9 tests we've discussed

Front 101

effect size

Back 101

it is the magnitude of effect (a continuous measure) that accompanies (not replaces) significance testing

Front 102

what is relative risk?

Back 102

ratio of the incidence of an event in two different groups

incidence of a smoker developing coronary heart disease is 60% and incidence of a nonsmoker developing coronary heart disease is 20% ...the incidence f a non-smoker developing coronary heart disease is RELATIVE to the same risk for nonsmokers

therefore 60/20 = 3 the relative risk is 3 times more likely that a smoker will develop CHD than a nonsmoker

Front 103

what is odds ratio?

Back 103

ratio of the odds - NOT the incidences

odds of an event occurring is the probability of the event divided by the probability of the event NOT occurring

Front 104

what does 1 in the odds ratio or the relative risk tell you?

Back 104

indicates no difference between the groups being compared

Front 105

main effect

Back 105

in factorial ANOVA it is the effect of one IV on the DV

Front 106

interaction effect

Back 106

the combined effect of two or more IVs on the DV

Front 107

standard error bars

Back 107

figures that show means should show standard error bars

read the figure caption because the lines could represent the standard error the mean, the 95% confidence interval, or sometimes the standard deviation

Front 108

survival curve

Back 108

shows the % of the group that has NOT had the event - assume that everyone is "good' and starts at 0 together

once you fall off the curve you cannot get back on the curve -

Front 109

alpha level

Back 109

significant level - the specified rate that we are willing to make a type I error; set by the experimenter - conventionally it is 0.05 or less

Front 110

power level

Back 110

1-beta = the probability that the researcher will observe a treatment effect when it occurs - will correctly reject the null; likelihood of finding statistically significant effect when it actually exists

want this to be high - this is the chance that you are correct

Front 111

effect size

Back 111

measure of the strength of relationship between two variables in statistical population - in the sample we are examining

Front 112

one tail

Back 112

directional - the hypothesis predicts an effect in ONE direction

Front 113

two tail

Back 113

nondirectional - the hypothesis predicts an effect but not the direction of the effect

Front 114

power

Back 114

determine the sample size needed to make sure your study has the ability to reject the null if the results show the smallest "meaningful difference"

Front 115

what is research?

Back 115

systematic study of a problem, issue or question -

done by comprehensive literature review, search in organized way for relationships between variables, or manipulating variables to see what happens to another variable

Front 116

what is the process of research?

Back 116

1-establish the need for info / research
2 - research questions and hypothesis
3 - research design
4 - sampling and manipulating data collection
5 - data analysis
6 - reporting and publication

Front 117

quantitative research

Back 117

done to answer a specific question by reliably measuring variable and using statistical techniques to examine the outcomes

begins with a specific hypothesis to examine a specific question

Front 118

what is the hourglass metaphor?

Back 118

begins with broad questions (research question or theory) which then gets narrowed down (research question)
operationalize, analyze the data, reach a conclusion (results from the study) and then generalize back to the questions (conclusions of the study)
then try to get generalized back to the broad questions

Front 119

qualitative research

Back 119

done to generate descriptive theory grounded in the data collected - no specific research questions to tackle and the researcher is the learner

provides the foundation and guidance for quantitative research with some exceptions

Front 120

what is a variable

Back 120

any observation that can take different values (gender, age, race, weight, self-esteem, blood pressure)

Front 121

what is an attribute

Back 121

a specific value of a variable

variable = gender
attribute = male and female

Front 122

what is a hypothesis?

Back 122

a specific statement of prediction - must be testable

Front 123

what are the two types of hypotheses

Back 123

null and alternative

Front 124

define a null hypothesis

Back 124

describes the possible outcomes other than the one you are conducting the study to examine

usually predicts the most boring thing - there will be no effect of the IV on the DV

Front 125

define the alternative hypothesis

Back 125

can be directional (one tailed) or nondirectional (two tailed) - a specific prediction that usually states what you expect to happen in your study

ie: the IV will affect the DV

noted as H1 or HA

Front 126

what is special about the null and alternative hypotheses?

Back 126

they have no overlap but together they account for all possible outcomes and are mutually exclusive and exhaustive

Front 127

what do histograms represent?

Back 127

probability! - graphic depiction of data

x axis - shows all values obtained

y axis - shows the frequency at which each value occurred

when the frequency is "normally distributed" it is graphically depicted as a bell shaped curve

Front 128

name 4 types of study designs

Back 128

1 - true experimental
2 - quasi experimental
3 - correlational study
4 - descriptive study

Front 129

what is a true experimental study?

Back 129

goal is to address a specific hypothesis by demonstrating that the manipulated variable is the only possible reason for the demonstrated outcome - most detailed and stenos design

IV if the manipulated variable -
CAUSALITY - afford causal statements - IV causes DV

Front 130

what 3 things are required of a true experimental study?

Back 130

1 - manipulation
2 - control
3 - randomization

Front 131

what is manipulation?

Back 131

researcher intervention - something the experimenter does to the subject -

treatment, intervention, prevention, medication

you must manipulate to endure internal validity - you are assured that what you are doing is causing the effect

Front 132

what is control?

Back 132

the experimenter's ability to control or eliminate alternative explanations for changes in the DV - be assured the changes were due to manipulation in the IV

Front 133

Threats to control

Back 133

time
biases by the participant or the researcher

Front 134

how does time threaten control of a study?

Back 134

would changes in the DV occur with our without the IV? control group - a group of participants that do not undergo the active component of the IV
Treatment as usual group

Front 135

how do biases affect threats to control?

Back 135

systematic effect that influences the DV other than the IV - not necessarily on purpose but it happens

Front 136

what are participant biases (there are 2 of them)

Back 136

demand bias
expectancy effects

Front 137

what is a demand bias

Back 137

features of an experiment which helps participants figure out what is expected of them - it can lead them to behave in a way that would satisfy those conditions

Hawthorne effect - the effect of the participants knowledge that they are part of a study and their performance level therefore improves -

effect in the expected direction but not for the expected reason

Front 138

what are expectancy effects

Back 138

occurs when a subject expects a given result and therefore unconsciously reports the expected result

Front 139

what is a placebo effect?

Back 139

positive reports on an outcome based purely on he belief that a "treatment" will work

Front 140

what is a nocebo effect?

Back 140

negative effect experienced from an inert substance due to the patient's expectations (opposite of the placebo effect)

think they got an active substance and now report awful side effect when actually they did not receive the medication

Front 141

what is an experimenter bias?

Back 141

one that the experimenter introduces and it affects the outcome

look more closely for expected effects in the treatment group so they find them

assess outcomes differently in experimental vs control group

unconsciously communicates desire outcomes to the participant - this increases participant's demand bias - you nonverbally and unconsciously communicate the results you want them to say

Front 142

how do you reduce threats to bias?

Back 142

double blind, single blind, open label or randomized control trial

Front 143

what is a double blind study?

Back 143

both investgiator and participant are kept ignorant of the group assignment

Front 144

what is a single blind study?

Back 144

a study in which either the experimenter or the subject is kept ignorant of group assignment but not both

Front 145

open label

Back 145

both the investigator and the participant know the participant is receiving the active treatment - mostly used exclusively in biomedical research - no inactive treatment would be given to participants

Front 146

what is a randomized controlled trial?

Back 146

GOLD STANDARD FOR BIOMEDICAL RESEARCH

Front 147

what is random assignment

Back 147

method of assigning participants to the different levels of the IV, so that every subject is the sample has an equal chance of being assigned to the experimental or control group

this is to maximize likelihood that groups are equalivant prior to the administration of the IV

Front 148

what is random selection

Back 148

how you draw the sample of people for your study

this would not be part of the true experimental study because people have to sign informed consent and select to be in the study - this is related to sampling

Front 149

what is a quasi experimental study?

Back 149

design in which the conditions of true experiments are approximated

this type of experiment is used when requirements of a true experiment are not possible or are unethical

Front 150

what is a correlation study?

Back 150

examines the naturally occurring relationship between 2 or more variables and provides information about direction and strength of the effect - but does not provide information about causality and temporality

can have no relationship, positive or negative relationships

Front 151

what are 3 types of descriptive studies?

Back 151

systematic review
comprehensive systematic review
meta analysis

Front 152

what is a systematic review?

Back 152

look what all studies find about a particular phenomenon - figure out what the literature says about a topic - now a study design but it does employ systematic methods

Front 153

what is a comprehensive systematic review?

Back 153

systematic review that intends to include all studies published that meet specified criteria - you cover everything published about a topic that is in the confines that you set

ie: everything published on MS in a patient 18-35 years old with early onset - you then look for EVERYTHING published and write a paper about your findings

Front 154

what is a meta analysis

Back 154

application of statistical methods to combine evidence from numerous studies -

HIGHEST LEVEL OF REVIEW

you do a comprehensive systematic review then compute a statistical value or effect size which is used to measure the size of the IV on the DV for each study - this is then averaged across all studies to give a more reliable picture of the effect

Front 155

what is the purpose of the methods section/

Back 155

allows the reader to understand exactly how you conducted the study -

how did you get the sample?
what instrument did you use?
was the study conducted in accordance to research ethics and what analytic plan did you use?

Front 156

what is the theoretical population

Back 156

who you want to generalize the data back to

Front 157

what is the study population

Back 157

what population you can get access to

Front 158

what is the sampling frame

Back 158

how can you get access to the study population

Front 159

what is the sample?

Back 159

who is in your study

Front 160

define population

Back 160

scientific group from which a researcher would like to represent his / her study - the group to which the results will be able to be applied to

Front 161

what is the sampling frame?

Back 161

the listing of the study population from which you will draw your sample tool

ie: phone book, pt records, customer list, etc.

Front 162

what is the sample

Back 162

the group of people whom you select to be in your study- all who sign informed consent - some that enter the study will not be compliant and will drop out and others will be compliant the entire time and do everything right - you can't just look at those who do everything right because this would generate bias

Front 163

sample bias

Back 163

sample that does not represent the true population - this results in a bad sample -

Front 164

what is oversampling?

Back 164

a process of including a larger representation in the sample than may exist in the population - this is intentional and purposeful sampling bias - - this is a balance because you can't force people in the study and you need to try to find the population that is most representative

Front 165

what is probability sampling?

Back 165

any method of drawing a sample that uses some form of random selection - it helps ensure that every individual in the population has an equal chance of vein chosen to be in the sample - the data from the study has the greatest generalizability - external validity

Front 166

what is simple random sampling?

Back 166

most basic form of probability sampling - common methods include table of random numbers, computer random number generator, etc...

likely to misrepresent subgroups

Front 167

what is non probability sampling?

Back 167

does not involve random sampling - less expensive and time consuming than probability sampling - the results are limited in external validity and there is no way of knowing that the population was represented well

Front 168

what is convenience sampling?

Back 168

the most common type of non probability sampling - the members of the population are chased based on ease of access and the traditional per on on the street or asking for volunteers

Front 169

what is ethics in research?

Back 169

active pursuit of balance between the right of the scientist to search for the truth and the rich of the participant to be protected from harm

Front 170

what is the nuremberg code

Back 170

developed in response to nazi docs that did experiments on concentration camp inmates without consent -
included rules about informed consent, right to withdraw, justification of scientific principles and beneficence

Front 171

what is the tuskegee syphillis study?

Back 171

studied the effect of untreated syphillis on AA men and when PCN became the standard treatment these men were denies the standard therapy

it is now unethical to withhold a treatment that is considered the standard of care

Front 172

what is the national research act?

Back 172

states that all institutions sponsoring research must investigate potential risks and benefits of every study conducted both in humans and animal

Front 173

what are IRB and IUCAC?

Back 173

IRB - monitors clinical research on humans

IUCAC - monitors animal research

Front 174

what is the belmont report

Back 174

document that provides ethical framework in US on the use of humans in experimental research - reference IRB nationwide

key principles:
respect for persons
beneficence
justice

Front 175

informed consent

Back 175

process - not an event - required before involving them in research

Front 176

3 key features of informed consent

Back 176

disclose information - tells participant everything in order to make and educate

facilitate understanding - researcher available to answer any questions

promote voluntariness - person decided whether to participate or not - once they sign the consent - they can also withdraw at any time

Front 177

informed consent should include

Back 177

purpose
procedurs
duration
risks / discomforts
possible benefits
alternative treatments
costs

Front 178

what is coercion?

Back 178

over or implicit threat of harm intentionally presented by one person to another in order to obtain compliance - bad thing to motivate participation

Front 179

what is undue influence?

Back 179

offer of an excessive or inappropriate reward to obtain compliance - promise a good thing in order to encourage participation

Front 180

what are psychometrics

Back 180

the field of study concerned with the theory and technique of educational and psychological measurement -

done on instruments NOT PEOPLE

Front 181

what is reliability

Back 181

repeatability or consistency - gives the same result over and over

necessary but not sufficient for validity

Front 182

what is test - retest reliability

Back 182

assess the consistency of a result from one time to another -

participant completes the instrument and the quality being measures should not change

same person will give the same answer at different times

Front 183

what is inter rater reliability?

Back 183

type of scoring agreement - used to assess the degree to which different observers give consistent estimates of the same phenomenon or observation

different assessors using the same instrument will produce the same answer

Front 184

what is parallel form reliability?

Back 184

used to assess the consistency of the results of 2 tests constructed in the same way form the same content domain -

same person gives comparable answers on related tests delivered sequentially -

would expect that you would get similar results on test A and test B

Front 185

what is internal consistency?

Back 185

deriving an estimate of how well items on the instrument that reflect these same overarching theme you are trying to get at and produce similar data -

many different tools used to examine an instrument's internal consistency - CRONBACH'S ALPHA is most common

Front 186

what is cronbach's alpha?

Back 186

average of all split half correlations - you have computed all split half estimates of reliability - you randomly divide items into 2 sets and calculate the total score from each set for each person and obtain the correlation coefficient to reflect the relationship between the sets - similar to parallel forms reliability but with split half you do not deliver the instrument in 2 forms

0.70 considered acceptable for showing the instrument has internal consistency

0.80 is considered excellent

Front 187

what are correlation coefficients?

Back 187

value that ranges from -1 to +1 and tells you both about the strength of the relationship between values and the patterns of the relationship

Front 188

what is validity?

Back 188

accuracy

Front 189

what are the 2 major types of validity?

Back 189

external
internal

Front 190

what is external validity?

Back 190

generalizability back to situation or group you are looking at this is with sampling portion - increase external validity when you draw a non biased sample

Front 191

what is internal validity?

Back 191

degree to which an instrument produces accurate and true data

Front 192

what is criterion related

Back 192

only possible when you have another instrument in existence that is acceptable as valid - examine the accuracy of the new instrument by looking at how its scores correlated with the results from another laid instrument

measuring against the GOLD STANDARD - criterion related is determined by comparing scores to an accepted operationalization of a construct rather than to the construct itself

Front 193

what is a construct

Back 193

the actual quality of interest, usually difficult to measure - intelligence, knowledge, anxiety

Front 194

what is operationalization

Back 194

the act of translating a construct into its result - actual test or instrument that you use

i.e.:beck anxiety inventory

Front 195

what is a criterion

Back 195

the result of successful operationalization - the criterion variable - scores on a block exam, scores on beck anxiety inventory

Front 196

what is predictive validity

Back 196

test's ability to predict something it should theoretically be able to predict - an instrument is supposed to measure some future performance then predictive validity should be investigated - comparison should be made between the instrument and some later behavior or score that it predicts

Front 197

what is concurrent validity

Back 197

test's ability to distinguish between groups that is should theoretically be able to distinguish between - occurs when the criterion measures are obtained at the same time as the new instrument score

Front 198

what is construct validity

Back 198

scores of the new instrument relate appropriately to a body of criteria that define a construct - approximate truth of the conclusion that your operationalization accurately reflects its construct -

you don't have a gold standard or you have a lot of other instruments so you have a group of number you able to compare your test to

Front 199

what is convergent

Back 199

degree to which the test is similar to other measures that it theoretically should be similar to - so when an instrument correlates highly with other variables with which it should theoretically correlate

Front 200

what is discriminant

Back 200

the degree to which the test is not similar to 9diverges from) other measures that is theoretically should not be similar to - instrument does not correlate significantly with the variable from which it should differ

Front 201

what are 4 levels of measurement

Back 201

nominal
ordinal
interval
ratio

Front 202

what is nominal data?

Back 202

the number simply names the attribute uniquely - no ordering rank is applied by the numbers - so they are arbitrary and mutually exclusive

Front 203

what is ordinal data?

Back 203

relationship between values reflect a rank order - degree of difference between the ranks cannot be determined

Front 204

what is interval data?

Back 204

difference between values reflects the attribute and does have meaning - the relationship between the attributes in known and exact - averages can be computed - intervals are equal utoriginate from some arbitrary origin

no absolute zero and no negative numbers

Front 205

what is ratio data?

Back 205

there is an absolute and natural zero point that is meaningful

Front 206

what are likert terms?

Back 206

ordinal metric and commonly used in survey research - these are level of agreement assessment 5-7 point scaled used most often and are generally treated as interval level data

Front 207

what is the data analytic plan

Back 207

it is a decision - it discloses how the data were treated, tools investigator used and special circumstances or treatment of data that warrant justification - if included is usually a paragraph in the methods section that tells the reader how you will analyze the data - should explain and justify the investigator's decision - if strong or suspect then it helps the critical reader interpret the results and conclusion that the authors present